JP6073680B2 - Submillimeter wave radar using phase information - Google Patents

Description

  The present invention relates to a signal processing device (signal processor) for a submillimeter wave radar system, this radar system, a robot or vehicle such as a road vehicle having this system, and a corresponding method and computer program.

  Discrimination between different objects in submillimeter radar currently relies exclusively on the measurement of the radar cross section, ie the energy fraction of the waves reflected by a reference unit square of an object. For example, based on the relationship between surface roughness and absorption characteristics, all points of the object exhibit similar radar cross-section values. Contour lines are formed by grouping similar measurements, which provides an identification tool for separating objects from the background. Submillimeter wave radar for outdoor applications faces complex objects in front of a complex background or behind a complex foreground. In typical conventional microwave radar applications (eg, ground-plane, plane-plane radar), this type of problem is because the background is empty, which is a substantially diffuse and low loss absorber. You will never encounter it. The only measured “sharp” radar cross section is created by the airplane as an object. Furthermore, the only measurable Doppler shift signal is derived from the angle at which the radar beam hits the ground and is provided by another airplane. Only low-flying airplanes remain undetected in the reflected signal from the ground.

  Microwave radars are generally not suitable for imaging applications that require target classification and certification in a complex environment. Several techniques have been proposed and demonstrated to enable measurements in at least moderately complex environments. One example is iceberg detection using naval microwave radar. Here, the small reflections of the iceberg (only a small part is found on the water surface) need to be distinguished from the larger reflections generated by the waves. Time-of-flight filtering can be used to determine the distance from the radar to the target, for example using frequency modulated continuous wave (FMCW) radar or chirped pulse radar. In addition, signal-sustained spatial frequency content needs to be analyzed (icebergs move much more slowly than embedded waves). Naval radar is effectively one-dimensional (depending on the structure of the environment) because most of the detection area is found in a small band below the horizontal line, thereby greatly reducing the amount of data to be processed.

  Road-running vehicles, such as passenger-car radars at lower frequencies (eg, vehicle-to-infrastructure 63 GHz, vehicle-to-vehicle 66 GHz, cruise control 77 GHz, collision prevention 79 GHz) are pulse radar structures or the above FMCW technology Either is used to identify objects at various distances. These radars do not generate images with sufficient resolution that can identify objects from their shapes. These radar systems provide significant horizontal resolution, but usually the vertical resolution is very limited.

  There are infrared cameras that provide high resolution in both the vertical and horizontal directions. Such a camera is sensitive to the total power (not coherent) absorbed by the pixel receiver, and in the case of a radar system, it cannot obtain distance information. Furthermore, most IR camera systems for automobiles are passive. Without sending a signal, the light emission of the scenario or the object itself is used as the signal source. Thus, IR camera systems need to deal with low contrast scenarios (humans standing in front of a warm background) and are sensitive to strong IR sources (lamps, sun) that blind the system. In addition, rain and snow substantially reduce the efficiency of the system.

  Submillimeter wave radar for outdoor applications must operate in a specific, well-defined frequency band. These bands are characterized by no atmospheric attenuation (or at least very weak). Radar above 100 GHz is available, as opposed to microwave radar applications where the full frequency range from 0 to 100 GHz can be used and only legal regulations limit frequency optimization Only a small part (20%) of the frequency range can be used. Typical bands (also referred to as “atmospheric windows”) are 100 GHz, 220 GHz, 350 GHz, 450 GHz 520 GHz, 600 GHz, 810 GHz, 1.000 THz, 1.650 THz and 2.400 THz with increasing peak attenuation. Located before and after. These windows are well known.

  WO 2006/075570 (HOLT) paragraphs 0063-0069, comprising irradiating at least a portion of a subject first with electromagnetic waves and then with different frequency ranges having a frequency of about 100 MHz to about 2 THz It is known to provide

  It is described that identification features, such as video patterns, intensity levels or colors, may be assigned to each video to enhance the observation of differences in the video. The video is a grayscale video generated from a color video. In the case described, the pixels in these images were assigned red, green and blue colors, respectively. This is at least partly the reason why the general intensity levels of these images are different.

  In addition, it is described that different videos generated from data for different frequency ranges can be displayed to the viewer continuously or simultaneously to help the viewer identify differences between videos. ing. Such areas of difference may correspond to video anomalies that may include objects. By directing the viewer's attention to the anomaly, identification of the object can be accelerated.

  The object of the present invention is to provide a novel signal processing device for a submillimeter wave radar system, such a radar system, a robot, or a robot vehicle or a road vehicle having this system, and a corresponding method and To provide a computer program.

  According to a first aspect, the present invention provides a signal processing apparatus for a submillimeter wave active radar system having a field of view, wherein the signal processing apparatus receives a signal received from the field of view and down-converted by the radar system. Configured to process, the downconverted signal corresponds to a predetermined pixel in the field of view having a time-varying amplitude and a phase component having a periodic component that depends on the content in the predetermined pixel in the field of view. Furthermore, the signal processing device provides a signal processing device configured to identify information about the content from the periodic component.

  By using phase rather than amplitude alone, there is additional information in the downconverted signal. This information has been lost without traditional non-coherence as indistinguishable from noise. Phase is more sensitive to changes in content such as objects, background and atmospheric conditions than to amplitude alone. Phase information allows for the retention of periodic components that can include at least three types of information specific to the content. First, content flutter in the form of quasi-periodic changes in content over a small number of submillimeter wavelengths, such as the movement of plants in the wind or other forced vibrations of a structure Can be specific to some content. Second, instead of being discarded as noise, changes in submillimeter standing waves can now be detected sufficiently to determine speckle differences caused by differences in content. Third, a thin layer on an object in the field of view can produce fringes that affect the received reflection of submillimeter radiation if that layer has a thickness on the order of multiple half wavelengths. it can. An example is an ice layer on a road. This additional information makes submillimeter wave radar more practical for various applications including, for example, automotive radar, which must not be affected by interference from sunlight, fog or rain.

  Embodiments of the invention may have additional features, some of which are described in the dependent claims and described in detail below.

  In some embodiments, the downconverted signal may be an IF signal having an IF spectrum, and the signal processing device is configured to separately analyze two or more regions of the IF spectrum. The This allows, for example, identification of content that preferably appears in one part of the spectrum but not in the other part.

  The signal processing device can be configured to perform any known technique in video analysis. Such techniques include digital filter applications such as smoothing filter or segmentation, contour generation, contrast jump, and / or filters used for line, edge, corner identification, contour, A filter adapted to use morphological operators to smooth regions and the like is included. The signal processing device may be configured to cluster portions of the received video, thereby identifying regions having common characteristics, for example, thereby identifying objects. This can include, for example, comparing the detected features with a library that stores features of a given object or object class.

  The signal processing apparatus can have a portion for associating a portion of the received signal with a corresponding set of receive beam positions. This allows related signals to be grouped into frames that represent the scene. This can be, for example, 1, 2 or 3D framing.

  The signal processing apparatus can have a portion for applying an operator to the framed data. This operator may be an operator for extracting local changes in amplitude and / or IF frequency in the scene at a given moment.

  This operator may be an operator for extracting a spatial or temporal feature in the framed data. This operator extracts characteristics that have a dependency on submillimeter frequency or amplitude, or that have a dependency on either the direction of radiation, the focal position of the receiver, or the polarization, or any combination thereof. May be an operator.

  In another aspect, a submillimeter wave having a transmitter that radiates to a field of view, a receiver that receives a signal from the field of view, and a demodulator that uses a local oscillator signal that is phase locked to the transmitter. A radar system, further comprising a signal processing device as described above, is provided.

  In another aspect, a corresponding method for using a transmitter and further using an active radar system that receives and processes signals from the field of view and a corresponding method for processing signals are provided.

  All additional configurations can be combined together and combined with all aspects. Other advantages will be apparent to those skilled in the art, particularly those skilled in other prior art. Many variations and modifications may be made without departing from the scope of the claims of the present invention. Accordingly, the form of the invention is illustrative only and is not intended to limit the scope of the invention.

1 is a schematic diagram of an embodiment of a radar system according to the present invention. 1 is a schematic diagram of an embodiment of a radar system according to the present invention. The figure which shows the step which concerns on an Example. The system figure of another Example. FIG. 5 is a schematic diagram of frequency and timing control for use in the embodiment of FIGS. 1, 2, 3 or 4 or other embodiments. FIG. 5 is a schematic diagram of frequency and timing control for use in the embodiment of FIGS. 1, 2, 3 or 4 or other embodiments. FIG. 5 is a schematic diagram of frequency and timing control for use in the embodiment of FIGS. 1, 2, 3 or 4 or other embodiments. FIG. 2 is a schematic diagram of an RF generator section for the same system embodiment. FIG. 3 is a schematic diagram of an LO generator section for the same system embodiment. 1 is a schematic diagram of an IF processing unit for the same system embodiment. FIG. 1 is a schematic diagram of an IF processing unit for the same system embodiment. FIG. 1 is a schematic diagram of a video processing unit for the same system embodiment. FIG. 1 is a schematic diagram of a video processing unit for the same system embodiment. FIG. FIG. 4 is a schematic diagram of an alternative embodiment using offline IF processing and video processing. FIG. 3 is a schematic diagram of beam shaping and scanning for an example radar system. FIG. 3 is a schematic diagram of beam shaping and scanning for an example radar system. A graph showing how content flutter and speckle effects can provide detectable information in the spectrum of a received signal. A graph showing how content flutter and speckle effects can provide detectable information in the spectrum of a received signal. A graph showing how content flutter and speckle effects can provide detectable information in the spectrum of a received signal. A graph showing how content flutter and speckle effects can provide detectable information in the spectrum of a received signal. A graph showing how content flutter and speckle effects can provide detectable information in the spectrum of a received signal. A graph showing how content flutter and speckle effects can provide detectable information in the spectrum of a received signal. A graph showing how content flutter and speckle effects can provide detectable information in the spectrum of a received signal. 6 is a graph showing a detectable interference effect in a received RF signal. 6 is a graph showing a detectable interference effect in a received RF signal.

  The present invention will be described with respect to certain embodiments and with reference to certain drawings but the invention is not limited thereto but only by the claims. The drawings described are only schematic and are non-limiting. In the drawings, the size of some of the elements are exaggerated for illustrative purposes and not to scale. Although the term “comprising” is used herein and in the claims, this does not exclude other elements or steps. When a limited or non-limiting item described as a singular noun is used, it includes a plurality of nouns unless specifically designated.

  The term “comprising”, used in the claims, should not be interpreted as being restricted to the means listed thereafter; it does not exclude other elements or steps. Thus, the expression “apparatus comprising means A and B” should not be limited to an apparatus consisting only of parts A and B. This means that for the present invention, the parts associated with the device are A and B.

  Further, the terms first, second, third, etc. in the specification and claims are used to distinguish between similar members and do not necessarily describe the order or time order. It is understood that the terms used in this manner are interchangeable under proper circumstances, and that the embodiments of the invention described herein can operate in a different order than those described or illustrated herein. Should be understood.

  Further, the terms upper, lower, upper, lower, etc. in the specification and claims are used for illustrative purposes and do not necessarily indicate their position relative to each other. Terms used in this way are interchangeable under the proper circumstances. It should be understood that embodiments of the invention described herein can operate in other arrangements than those described or illustrated herein.

  It should be noted that the term “comprising”, used in the claims, should not be construed as limited to the means listed thereafter. This does not exclude other elements or steps. Accordingly, it should be construed as specifying the presence of the described configuration, integer, step or part, and the presence of one or more other configurations, integers, steps or parts, or groups thereof. It is not excluded. Therefore, the scope of the expression “device with means A and B” should not be limited to devices consisting only of parts A and B. In the context of the present invention, it is meant that the components only relevant to this device are A and B.

  Throughout this specification, references to "one embodiment" or "an embodiment" include certain configurations, structures or features described in connection with the embodiment that are included in at least one embodiment of the invention. Means that Thus, the appearances of the phrases “in one embodiment” or “in an embodiment” in various places throughout this specification are not necessarily all referring to the same embodiment, or are so. May be. Furthermore, certain configurations, structures or features may be combined in any suitable manner in one or more embodiments as will be apparent to those skilled in the art from this disclosure.

  Similarly, in describing exemplary embodiments of the present invention, various configurations of the present invention are often described in one embodiment, drawing or description thereof, in order to streamline the disclosure and in one or more of the various configurations. Grouped together to help understand the inventive aspects of This method of disclosure, however, should not be interpreted as reflecting an intention that the claimed invention requires more features than are expressly recited in each claim. Rather, as the following claims reflect, inventive aspects exist in less than all configurations of a single disclosed embodiment described above. Thus, the claims following the detailed description are hereby expressly incorporated into this detailed description, with each claim standing on its own as a separate embodiment of this invention.

  Further, some embodiments described herein may include some configurations included in other embodiments, but not other configurations included in other embodiments, while combinations of configurations of different embodiments. Are intended to be within the scope of the present invention and, as will be appreciated by those skilled in the art, form different embodiments. For example, in the following claims, all claimed embodiments can be used in all combinations.

  In the description presented herein, numerous specific details are set forth. However, it should be understood that embodiments of the invention may be practiced without these specific details. In other instances, well-known methods, structures and techniques have not been shown in detail in order not to obscure the understanding of this description.

  The invention will now be described by a detailed description of several embodiments of the invention. Obviously, other embodiments of the invention may be constructed based on the knowledge of those skilled in the art without departing from the technical disclosure of the invention, and the invention is limited only by the terms of the appended claims. .

  “Active” means any radar system that emits radiation to illuminate the scene and detects the radiation reflected from the scene. If the reception is phase locked with respect to the radiator, the radiator can in principle be independent of the receiver by detecting the radiation. This is in contrast to “passive”. By passive is meant any radioactive system that uses either a broadband source (light source) or the target's own radiation to illuminate the target (either human made or not). Such a system is considered to be “passive” radar by those skilled in the art.

  Sub-millimeter radars are generally intended to include all radars that use frequencies above about 100 GHz, and the examples are within a narrow range below 300 THz, also known as terahertz radar, below 3 THz. Explained. Such radar can be applied, for example, to systems for vehicles and, for example, security or surveillance systems in buildings as is well known.

  Content is defined as everything in the field of view, for example, an object to be detected such as a car or a pedestrian for automotive applications, and at least a noise for automotive applications. And include backgrounds or obscure objects such as plants or buildings that can be deleted from the video as part of the identification. The content further includes atmospheric effects that affect either or both external illumination and reflections returning to the receiver, such as rain, fog, snow or sunlight.

  “Vehicle” should be interpreted broadly and can refer to all robots, robot vehicles, self-guided vehicles, road vehicles, ships, airplanes, and the like.

Example Introduction: Challenges and System Considerations Submillimeter radar (hereafter submillimeter) radar for outdoor applications faces complex objects in front of a complex background, unlike other radar systems. Therefore, new technologies are needed to enhance radar identification and classification capabilities without undue bandwidth and power requirements.

  Reducing RF bandwidth is one useful way to improve noise performance. Classic Doppler and FMCW radar systems require a very large RF bandwidth to reach the speed required for their respective round trip distance resolution. The only possible way to reach an acceptable signal to noise ratio is integration. As a result, only a few measurements can be performed per unit time. This number is usually sufficient for 1D scanning radar. Depending on the number of pixel measurements required, an image radar system cannot be realized based on these classical methods. There should be a way to reduce the required RF bandwidth and provide classification information in other ways. This reduction in bandwidth results in a better signal to noise ratio. This increase in signal to noise ratio can be used to reduce LO power requirements or to increase the number of pixels (frames) measured per unit time.

  LO power is another limiting factor in making submillimeter image radar today and is the availability of RF power to illuminate a given distance scenario. At present, the output of 10 μW is the highest level using a waveguide block. For an affordable radar system, an output of 10 mW must be achieved. By adding broadband illumination requirements, the radiant power target becomes even more difficult to achieve. A way to reach a reasonable output is to build a multi-frequency source away from the broadband source. This multi-frequency source radiates the required power in the required frequency band, but need not be capable of producing high power levels in the frequency range between the atmospheric windows.

  Since active submillimeter radars are usually sensitive to reflective surfaces at or near specular angles, the shape of an object that is “visible” at submillimeter frequencies is essentially the same as the image acquired using an IR camera. Is different. This video is also different from the video acquired using passive millimeter-wave radar under outdoor conditions. Under these conditions, the essentially cold sky provides a large image contrast to the warm ground. By using a passive submillimeter radar at a frequency of 200 GHz or higher, the sky temperature is found to be close to the ground temperature, resulting in a reduction in image contrast towards indoor conditions. The face image in the active submillimeter wave radar mainly consists of bright reflections of the nose tip, forehead and chin. Since this radar system provides illumination, there is no need to consider indoor and outdoor conditions. The rest of the face does not contribute to the image because they are not in the specular angle region, but there is very little diffuse reflection, unlike the optical frequency. Another factor limiting the diffuse reflection contribution to the image is the limited dynamic range of the receiver. Therefore, the shape of the object does not correspond to a recognizable shape in the optical image, and therefore another level of information needs to be added to classify the object. Knowing that almost every point on the face will move at the same pace, the RF speckle pattern frequency spectrum will be the same, even if the absolute level of the signal is low.

  An object classification library can be a useful part of the identification process. Along with familiar multi-frequency radar systems, a library of object behavior must be pre-configured. Not only the specific behavior of the object (with respect to the RF frequency), but also the low frequency correlation length (measured in the IF band) can be listed to identify different object classes.

  In addition, the object classification library can further include information about the spatial and temporal correlation functions of the object. Some objects have a typical size and shape on the imaging device (spatial correlation length) and a typical visible time (correlation time), as well as relative movements of object parts relative to each other (object autocorrelation spectrum). Have.

  Contrary to conventional laser systems, submillimeter wave automotive radar can be used as an imaging device. Thus, a new set of concepts can be developed: image radar requires high pixel recall and small integration time at each pixel. Therefore, and with high spatial resolution obtained based on smaller wavelengths, speckle signatures caused by small movements of the object and background are visible in the amplitude information of the IF signal. They appear as noise unless phase information is maintained and detected. Some conditions are as follows.

Spatial resolution must be on the order of moving objects (leaves, clothing, vehicle parts).
The wavelength must be small enough so that such complex object movements cause large changes in the phase information.

  Both of these conditions are satisfied with submillimeter wave radar, unlike other types of radar. Spatial resolution of a few centimeters allows identification of individual leaves, clothing parts and vehicle parts. Usually, wavelengths smaller than millimeters result in small movements on the order of a fraction of millimeters and become visible as shifted standing wave rates, ie classic speckle patterns.

  Instead of removing these periodic components such as speckle patterns, their frequency content can be evaluated. Due to the typical frequency range of the microscopic movement of the object under investigation, it leads to the availability of a noticeable micro-motion spectrum from object class to object class.

  The leaves of a tree usually have a vibration spectrum of several mm / s leading from a speckle frequency of 10 s to 100 Hz. The garment has a vibration spectrum of a few mm / s resulting in a sub-Hz spectrum. Conventional radar systems have a frame rate of several tens of Hz and cannot see these speckle patterns. By performing a series of measurements on selected pixels over a time window, the frequency resolution can be increased to a few kHz. By taking a few pixels of information and performing an FFT analysis or other transformation, it is possible to remove certain noise, such as plant effects on the car case.

Many more are performed, as described in more detail below.
Correlation analysis can add two more dimensions to the 2D video obtained by single frequency radar.

  The correlation time for all given pixels can be calculated “back in the past” and the correlation length for all adjacent pixels can be calculated “simultaneously”. The total output of this method is apparent by calculating the correlation of any pixel with respect to other pixels taken at the previous time. This “total” correlation includes time and space-like characteristics such as object movement relative to the scenario frame, movement of the object part relative to other parts of the object, object size and persistence information. All of these characteristics are content specific and allow the identification and classification of various types of objects.

  For an imaging system, the frame rate should be chosen fast enough so that both the background and object correlation times are slower than the frame rate. Calculation of the correlation (autocorrelation time) of each pixel point with itself “back in time” produces a contrast enhanced image. This object usually has a longer correlation time than plants in the background due to small movements caused by the wind. An object (eg, a pedestrian) typically has a correlation time that is much faster than a piece of road facility or a human cardboard image used for advertising, for example.

  By taking the same measurement information and calculating the correlation of pixels to adjacent pixels, a typical correlation length for a particular class of objects is obtained. Plants have small, comparable correlation lengths caused by small movements, humans have longer correlation lengths that depend on their size, and road facilities often have the longest correlation length depending on their size .

  Next, calculating the time and space correlations, two new dimensions were added to the measurement data. That is, not only the size of the object, but also the duration and typical movement of the object relative to the radar system can be used as a detection tool.

  These “full” correlation spectra also contain velocity information (perpendicular to the radar system) with distinct correlation peaks associated with pixels where the object is visible in a continuous set of frames. The velocity information in the direction of the radar beam can be obtained in various ways by identifying the displacement or the size of the target when correlating the frame with the previous one.

Because of the relatively large frequency distance between the plurality of submillimeter band used usable submillimeter band, radar cross-section of a given obstacle is to cause the physical properties of the material of the obstacle, different frequency bands So it will be different. The human body reflects all submillimeter waves in much the same way. That is, the signal loss caused by human skin acts as an almost perfect mirror independent of the radiation frequency. Normal tree leaves with a thickness of 0.5 mm, ... 1 mm show a Fabry-Perot resonance effect, and the thickness of the tree leaves projected perpendicular to the radar beam is an odd number, half the wavelength in the material If it is twice, it will result in destructive interference. At other frequencies, the same tree leaf has a thickness that matches an even multiple of half the wavelength in the material, resulting in constructive interference. The destructive interference results in a radar cross section of zero, thus making the leaves invisible. The constructive interference frequency maximizes the radar cross section.

  In order to make effective use of these characteristics, a multi-frequency radar system can be used to evaluate the acquired image coherently. Selecting one or more (for example, three) frequency bands for each of the radar images obtained within a certain band is a predetermined attribute of “THZ color”. Thereafter, all the images ("extract THz color") obtained at a certain time from a specific scenario (scene) are overlapped to form a false color image. As a result, the frequency dependence of the radar response is converted into a color image more suitable for image processing. The frequency band associated with color results in a “white” image from the human body and a “colorful” echo from the leaves. This is because grass leaves have a higher frequency and destructive interference first occurs compared to tree leaves. Tree leaves then have a higher frequency than coniferous leaves. The cardboard walls and traffic signs have a distinct non-white color, which does not change at high speed in most of the image (much like an oil spot on the surface of the water). The black ice layer on the pavement also has a slowly changing appearance of typical colors. In contrast, the snow is completely black. This is because multiple reflections on the snow surface completely absorb the incident radiation.

  Therefore, false color submillimeter waves and THz imaging serve to discriminate objects based on the physical surface structure of the objects. The total signal amplitude obtained in all frequency bands ultimately symbolizes the radar cross section. In this data set, the discrimination parameter can be mapped based on “THZ color”. For example, three frequency band images can be taken. The “red” color essence is acquired at 350 GHz, “green” is acquired at 450 GHz, and “blue” is acquired at 500 GHz. Human echo is common to all frequency bands. The echoes of leaves and shrubs are strongly frequency dependent. By superimposing these three color images, a false color image is generated as a result. Here, the human echo is “white”, and the leaf echo is colored. Therefore, in a coherent radar system, multi-band imaging can be used to enhance contrast and allow object classification.

  The submillimeter wave image radar processes a scene (scenario) that is much larger than the wavelength of radiation used by all of the imaged parts. This is the opposite of conventional radar, where a typical imaging object has a typical size of several wavelengths, so only direct movement can be measured in phase Changes. As a result, submillimeter wave radar images contain more details of the surface properties of the imaging object than conventional radars. On the surface of a natural object, a typical structure with a characteristic length of a few millimeters is often found. This includes bark, leaves, clothing, and reflective surfaces on traffic signs. Therefore, the radar cross section encountered at a certain point on the obstacle depends more strongly on the radar image used in the narrow band path than the microwave radar image. Extracting and handling this information means that it can be used to detect and classify objects or obstacles.

  As mentioned above, the maximum information from the radar signal is the spatial spectrum (how large and how small the object is), the low frequency spectrum (how fast the small part of the object moves relative to the object), IF Variations in periodic components such as frequency spectrum (how fast the entire object moves relative to the radar system) and color, RF frequency spectrum (how the reflectance of the object changes with THz frequency) Can be extracted.

  Considering this information and previously acquired frames provides a fast object discrimination tool for the radar system. With such a tool, it is possible to avoid reaching a continuous wave radar with information comparable to Doppler radar, for example using the IF frequency spectrum, which has the most negative impact on system noise.

  All coherent radar systems that receive reflections are limited by their noise equivalent reflectance difference (NERD). This NERD is more suitable in this frequency range than the noise equivalent power used in traditional radar systems. A reflectance difference smaller than this cannot be detected. Decreasing NERD is extremely important to strengthen the actual radar system.

There are two fundamentally different strategies to enhance NERD.
1: To increase the difference in the measurement signal obtained at a given reflectance difference. By operating on a frequency band set instead of operating on a single frequency, the difference in signals received with a given reflectance difference is enhanced assuming that the reflectance is frequency dependent.
2: To reduce the noise of the measurement system. By operating on a set of frequency bands instead of operating on a single frequency, the resulting amount of data is doubled, thus integrating the measured data volume more than twice and reducing noise. Substantial reduction to allow object identification. This is because noise tends to be canceled by being more integrated.

  By taking into account not only the noise of the measurement system but also the noise caused by rain and snow, this noise can be reduced by measurements on the frequency set. That is, only the object is correlated in the measurement. Rain scattering and absorption is strongly random and frequency dependent.

  Operating on a set of frequency bands substantially eliminates the speckle effect. Speckle is a notorious by-product of all coherent radar systems caused by standing waves between transmitter, object and receiver. Speckle filtering usually results in significant losses in the signal-to-noise ratio and the measurement signal dynamics. Multiple radar systems do not require any speckle filter when operating at a sufficiently large number of frequencies.

Use false color components of multiple settings within the same submillimeter wave band can also be obtained, for example, by changing pixel resolution, integration time, and IF bin settings of video taken at different settings and comparison video . From this, more accurate Doppler information is available at several points in the video. After contour line extraction performed at a higher resolution, this velocity information is allocated to the entire object delimited by the high resolution contour lines.

System components: Tx units Multicolor THz radars require multiple sets of frequency multipliers or one multiplier set that can generate large output power for object illumination within a frequency band set. Suitable frequency bands for automotive applications are 350 GHz, 450 GHz, and 515 GHz, as determined by an atmospheric window with low enough water absorption. Continuous wave transmitters are more practical when phase lock is used. Because any pulse response can be calculated based on the CW result obtained using the SAR technique. Furthermore, it is technically much more difficult to recover phase information from pulsed transmitter data at submillimeter frequencies.

System parts: Rx unit A subharmonic mixer stage is suitable for the receiver side of multicolor THz radar. The required LO frequency is typically only 2 or 3 times lower, and the power available to pump the mixer can be generated more easily. Different color channels can be separated using a Martin-Puplett diplexer stage as an optical beam splitter. Here, almost 80% of the input multi-frequency signal can be used compared to an absorptive solution where only 30% of the signal can be used to generate the measurement signal. By dividing the LO and FR sections into three independent units, the total THz radiated power is increased by a factor of 3, thereby increasing the signal to noise of the system by the square root of 3, ie 1.7. .

  Another solution is to use a sideband generator based on a single sideband upconverter on each RF source signal for each RF band. With different sideband displacements for each RF band, only a conventional mixer is required and the various RF bands are included in the IF signal by frequency multiplexing.

  Again, the Rx unit requires multiple sets of frequency multipliers or a single set of multipliers that can generate significant output power for object illumination within one set of frequency bands. Suitable frequency bands for the subharmonic mix are 175 GHz, 225 GHz and 257.5 GHz. None of these frequency bands are harmonics of each other, so RF crosstalk is not a problem.

  Since 2D video radar systems have very difficult requirements on the available instantaneous IF bandwidth, the signal-to-noise ratio is a problem. (The constraints on the RF bandwidth are more lenient.) This can only be satisfied by simultaneous measurements in the THz frequency set, thereby avoiding the use of bandwidth increase modulation techniques. On the other hand, simultaneous measurement at one set of frequencies multiplies the amount of orthogonal data obtained to further reduce system noise.

  Considering the RF frequency dependence of the reflectance of the object, more information is obtained than with conventional laser systems, thus generating further classification power. By using a typical library of RF frequency spectra, it is possible to classify radar objects. In particular, plants (leafs) have a thickness on the order of THz wavelengths. Therefore, these echoes are random in frequency. Any correlation analysis between several frequency channels effectively removes all echoes from the leaves.

  Further resistance to rain and snow can be obtained by several independent frequency channels with object information only as a coherent base. Any additional atmospheric noise and rain / snow noise behave as random reflections (in time and frequency). Thus, the correlation function between frequency channel pairs effectively removes snow and rain from the radar signal. Traditional (consumer) radar systems have bandwidths that are very limited by legal regulations. This does not apply to submillimeter wave radar. The complete frequency space can be used to reduce noise.

  In all coherent radar systems, speckle effects caused by standing waves between the transmitter, the object and the receiving antenna can be reduced. This standing wave pattern is extremely frequency dependent. Therefore, by operating on several frequency channels, all undesirable speckle effects can be substantially eliminated and no spec-removal filter is required.

1 and 2: Schematic diagram of an embodiment of a radar system FIG . 1 shows a portion of a system that includes a transmitter and receiver portion 10 and a coherent demodulator 20 that is phase locked to the transmitter. The system forms a down-converted signal having a time-varying phase and an amplitude component including a periodic component that depends on the content at that point in the radar system's field of view for a given pixel. In principle, the receiver and transmitter can be of all types, for example scanning or otherwise 1D, 2D or 3D scanning. The range can be any range suitable for the application, so automotive applications may require a 100 meter range or higher, while medical scanning or material analysis may be several meters or 1 It can be in the range of meters or less. Phase locking can be performed in various ways, some examples of which are described in detail below.

  The signal processing unit 30 can identify information related to the content from the phase and amplitude information in the down-converted signal. Signal processing can be performed in any way, offline or in real time, in software for a general purpose processor, or in other conventional techniques suitable for cost, speed, power consumption and other categories for each application.

FIG. 2 shows another embodiment, which is similar to FIG. 1, but shows one implementation of each portion of FIG. 1 in more detail. The transmitter illuminates the object 70 and its reflection is received by the receiver 90. Multiple narrowband scans are performed in different submillimeter wave bands, denoted as RF ₁ , RF ₂ and RF ₃ . Demodulator 110 is phase locked to the transmitter by a local oscillator 100 that provides signals to the transmitter and demodulator. The demodulator can selectively demodulate the various bands and output an intermediate frequency IF signal, for example, as a parallel stream or in series in successive lines or columns or frames.

  The spectrum analyzer 150 outputs a number of spectra for each pixel for a selected portion of the entire IF band. This spectrum may undergo a frequency transformation or other transformation. For the spectrum, three “bins” are shown, where there can be thousands or tens of thousands of pixels, some of which are shown for clarity. Spectrum 150 is for RF band 1 and IF band 1. Spectrum 160 is for RF band 1 and IF band 2. Spectrum 170 is for RF band 2 and IF band 1, and so on.

  Portion 180 processes these spectra to extract content specific features caused by content swings, speckle patterns and surface layer interference effects. The raw results of the shapes of the plurality of contour maps are stored in the storage device 220 and subjected to object detection and classification by the portion 200. This can compare their features or thresholds according to the object model library 200. Desired content objects are enhanced and unwanted noise (reflected waves) is reduced or eliminated. As a result, the video 230 shows content.

  FIG. 3 shows the operational steps of another embodiment starting with a scanning step 240 for scanning the field of view FOV. The received signal is then coherently demodulated at step 250. In step 260, the down-converted signal is transformed from the time domain to produce a frequency or similar spectrum, highlighting changes in periodic components. In step 270, content-specific features are extracted, for example, by various types of correlations. The object is classified by comparing it with the model at step 290, and its features and classification are output at step 300 and used for all purposes. Some examples include video output, alarms, and control signals that assist the driver in controlling the vehicle.

FIG. 4: System according to another embodiment FIG . 4 shows a submillimeter wave system with a transmission chain on the top line. This includes an RF generator 310 that receives a variety of different oscillation frequencies on the order of 11-14 GHz for different submillimeter wave bands and a driving signal of 350 to 515 GHz to drive the transmit antenna 315. Is configured to output. Transmit antenna 315 may be, for example, a horn antenna or other antennas known to those skilled in the art. Portion 320 directs and scans the transmission beam to cover the field of view. In principle, the transmit or receive beam can be “steering”, ie a wide-angle beam that is not scanned.

  Portion 340 shapes and scans the receive beam on the FOV. The receiving antenna 335 down-converts the signal received at 350 to 515 GHz and supplies it to the mixer 345. This mixer may be a subharmonic mixer and is supplied with a local oscillation signal from the LO generator section 330. This local oscillation signal is phase locked by the frequency and timing control portion 325 to the same oscillation signal used at the transmitter. The LO generator part can take in a phase-locked oscillation signal from 11 to 14 GHz and output a frequency from 175 to 258 GHz to the mixer. This mixer can output a down-converted IF signal in the range of 0 to 11.8 GHz to the IF processing portion 350. This is supplied to the video processing portion 355, which supplies signals to the user interface or further processing portion.

FIGS. 5, 6 and 7: Schematic diagram of frequency and timing control FIG . 5 shows the frequency and timing control portion of FIG. 4 for use in the system embodiment of FIG. 1, 2, 3 or 4 or other embodiments. An example of how to implement 325 is shown. A common or reference oscillation signal is supplied from a stable source to a partial M / N phase-locked loop 375, which is a signal having a frequency M / N of the stable source and having a fixed phase relationship with the stable source. Output. This output is branched and one path is directed to the RF oscillating portion 310 via the other phase-locked loop M1 / N1 · 380 to provide a 11 to 14 GHz signal. The other path is directed to LO generator 330 via another phase-locked loop generator M2 / N2 · 385 and provides a 11 to 14 GHz signal. This signal is in a fixed phase relationship with the common source and the signal supplied to the transmitter in order to produce coherent demodulation. The signal going to the LO generator is at a slightly different frequency, so the difference creates a mixer output band. This difference depends on the selection of the values N1, M1, N2, M2, and the step size at which the difference changes depends on N and M. A control computer for the radar system, not shown, can be used to set or change these values. Portion 375 sets the steps that can change the difference between the LO and RF outputs. The actual RF frequency is the product of parts 375 and 380, and LO is the product of parts 375 and 385. The choice of M, N gives a scale that can measure the frequency difference between the output 380 and 385 of the frequency generator. (This M / N * Fosc replaces the normally fixed quartz standard used in PLLs and allows for a more flexible and programmable RF band structure.) Frequency transmitted to 310 and 330 Therefore,
M * M1 / (N * N1) * Fosc and M * M2 / (N * N2) * Fosc
Given by.

The final frequency generated by the LO and RF sources is then:
M * M1 / (N * N1) * Q * Fosc and M * M2 / (N * N2) * Q * Fosc
Determined by.
Here the input frequency is Fosc and the factor Q is given by a fixed hardware structure in the frequency multiplier block.

  Since the “phase check time” is not tuned as much as possible, selecting the factors M, N, M1, N1, M2, and N2 as a pair of prime numbers can reduce phase noise in a very efficient manner. it can. In addition, this avoids unrealistic responses caused by signal crosstalk at the same or subharmonic frequencies. This configuration is particularly interesting. This is useful for calculating IF correlation over a large IF range when at least some major frequencies (used in the PLL above) are used as IF frequencies. Main frequency M * M1 / (N * N1) * Fosc, M * M2 / (N * N2) * Q * Fosc, M * M1 / (N * N1) * Q * Fosc, M * M2 / (N * N2) ) Moving * Fosc and M / N * Fosc out of the analysis IF region increases the dynamic range of the system.

  FIGS. 6 and 7 show alternative embodiments for implementing a phase locked loop generator 380 or 385 for use in FIG. FIG. 6 shows a phase locked loop. The frequency divider 400 divides the input frequency by a factor N. This is supplied to one input of the phase comparator 405. Its output is a phase error, which is provided to tune the voltage controlled oscillator 410. This output is fed to frequency divider 415 to divide its frequency by factor M and feed it to the other input of the phase comparator. This means that the output of the VCO 410 is stable and has the phase lock frequency M / N of the input signal. M and N should be chosen so that they do not have a common factor and are prime to each other so that they do not lock together. Since individual power is supplied to different parts and many update points are spread equally over time, separate power should be provided to maintain a good phase relationship between LO and transmitter .

  FIG. 7 shows an alternative example that has a similar configuration to the phase-locked loop of FIG. 6, but the frequency multiplier 411 after the VCO 410 is part of the control loop. This multiplier multiplies the frequency by a factor Q, which in one example is 36. The output of the multiplier is the total output, which in turn is a Fabry-Perot “sharp” bandpass filter 412, a power detector 413, a part for detecting the ratio of filtered and unfiltered power. It is fed back to the second input of the phase comparator via 414 and the subsequent chain of VCOs 416. A second power detector 417 is provided to detect power before the band pass filter. The overall output is a phase locked signal at the input frequency QM / N. This substantially incorporates several multiplication stages of the RF or LO generator into the fractional frequency generator. This helps reduce the phase error introduced by the multiplier, but at the cost of making a more complex fractional frequency generator. In FIG. 6 or 7, the values M and N are programmable by the control computer, and when using a single frequency radar where the frames need to be taken continuously at different RF frequencies, which part of the submillimeter wave band is currently To determine if it is used in the video. For a true multicolor system, several transmitters and receivers are necessary to obtain images simultaneously at different RF frequencies.

  The actual values of M and N are programmable by the radar system's control computer, receiving a recipe for the measurement settings for a particular frame. A “simple” phase-locked loop (PLL) structure can be seen in FIG. Start with phase comparator 405. This phase comparator generates a signal proportional to the phase difference of the input signal. It therefore produces a constant voltage when the signals at both inputs have the same frequency and thus a constant phase difference. If the frequencies of both inputs are different, the phase difference at the input will increase (or decrease) linearly with time (in fact, the phase difference modulo 2π will result in a sawtooth signal). . By using a filter at this phase detector output, the average value of the phase detector output is positive (negative) when the lower input channel carries a lower (higher) frequency than the higher channel. It is necessary to confirm. By supplying this signal to a voltage controlled oscillator (VCO) 410, a large (small) input voltage results in a high (low) frequency, which is then divided by a factor M that can be programmed using divider 415. Later), the frequency increases (decreases) on the low input side of the phase detector 405, and the frequency mismatch found at the first location is substantially corrected. When stable, the frequency generated by VCO 410 is M times greater than the frequency found at the top input of phase detector 405. This upper input frequency is itself given by the input frequency Fin divided by N using the frequency divider 400. Thus, under the assumption that the PLL is locked, i.e. the output frequency of the phase detector is smooth and does not include a sawtooth row for a longer time, the output frequency is M / N * Fin. The frequency at the output M / N * Fin is limited to tens of GHz (using today's frequency divider technology). In order to create a submillimeter wave frequency, it is necessary to add a fixed multiplier chain with a factor Q to the output of this PLL. This makes the phase noise generated by the multiplier “out of the loop” and therefore not attenuated by the PLL filter.

  The alternative method of FIG. 7 is essentially a frequency locked loop (FLL). Because the phase is not directly locked, it is only indirectly locked through frequency locking. It should be noted that when such an FLL generator configuration is used, the phase relationship between the input frequency and the output frequency slips slowly over time. Therefore, it is necessary to prepare a reference channel, and it is necessary to double the down converter stage. As a result, the reference phase and the measurement phase become available and the difference creates a phase delay to the target. Here, the output of VCO 410 is multiplied by factor Q in multiplier 411. This output signal can be analyzed using a sharp bandpass filter centered outside the RF frequency interval of interest. When using a power detector 417 to compare the signal from a sharp bandpass filter to the overall output, the ratio between the “sharp” and “wide” or unfiltered output increases with increasing RF frequency. Generate voltage. This frequency dependent voltage is delivered to the second VCO 414, where a lower reference frequency is generated, which closes the loop in a manner similar to that shown in FIG.

  There are a series of basic differences. Using a low-pass filtered variable that depends only on the total power absorbed (at 413) destroys the phase relationship between the input and output frequencies.

FIGS. 8 and 9: RF Generator Portion and LO Generator Portion System Embodiment FIG. 8 shows a possible implementation of the RF generator portion 310 for use in FIG. 4 or other embodiments. The phase locked oscillator signal is supplied to a 12 GHz input amplifier 450. This is fed to an x2x2 multiplier and amplifier 455 and multiplier 460, connected in series with a 10-16 GHz signal, to provide a further frequency multiplication of x2 in three stages. Finally, the waveguide filter 465 acquires the 320-512 GHz signal and applies a cutoff to 350 GHz. Other values can be used to generate other submillimeter wave bands.

  FIG. 9 shows a schematic diagram of the LO generator part, which has a corresponding part in a similar chain, but half the frequency that is suitable when a subharmonic mixer is used. End output. A phase locked oscillator signal is provided to a 12 GHz input amplifier 480. This provides a 10-16 GHz signal to an x2x2 multiplier and amplifier 485 and multiplier 490 connected in series to achieve further x2 two-stage frequency multiplication. Finally, waveguide filter 495 obtains the 160-256 GHz signal and applies a 175 GHz cutoff. Other values can be used to demodulate other submillimeter wave bands.

10 and 11: IF processor 350 for the same system embodiment
FIG. 10 shows an implementation that includes a LO2 generator 500 that feeds an IQ mixer 505. The mixer's second input is similarly supplied with a signal from the mixer 345 via an input amplifier 515 and an input filter 510 in series therewith. The output of the IQ mixer is a 65 MHz composite signal which is supplied to an IF2 processor 520 whose details are shown in FIG. The parts shown in FIGS. 10 and 11 can be implemented in the digital domain if the sampler was included at the appropriate point in the circuit.

  FIG. 11 shows a possible implementation of the IF2 processor 520. The input is supplied to a 65 MHz range selector 560 and then supplied to another IQ mixer 580 via an input filter 555. The other input to the mixer is from the LO2 generator 550, which provides a signal in the 0-65 MHz range. This mixer outputs a signal having an output range (range) of 44.1 kHz to a finite response filter 565 used for noise reduction by decimation and smoothing. An example of a suitable integrated circuit is the Analog Devices AD6620 chip. This powers the finite response filter 570 required to achieve full quadrature demodulation of the input signal, and then powers the autocorrelator 575 used for data decimation and noise suppression. This configuration is very traditional and is used, for example, in GSM mobile stations. The purpose of data decimation is to increase the practical number of bits provided by the AD converter. By using this triple filter structure, a 12-bit converter can physically generate 22-bit composite data. The third of these FIR filters is the upper and lower sidebands as soon as the sidebands are correlated (the echo has a net frequency displacement of zero in small plant movements). It can also be used to decorrelate the signal.

FIGS. 12 and 13: Video Processor Implementation FIGS. 12 and 13 show the implementation of the video processor 355 for the system of FIG. 4 or other embodiments. The unprocessed IF signal from the IF2 processor is a frequency domain signal sequence that represents the spectrum at each pixel of the sequence. The input to the frame generator is also another signal sequence related to the time to the position of all (transmitter and receiver) antenna beams. The frame generator generates this relationship between antenna beam position and time, effectively generating beam position data and (2 + n) D video (under the assumption that the beam position sweeps the 2D area over time. Used to assign to the IF data stream. N represents the number of additional dimensions. For example, n = 1 results in each image pixel consisting of an IF spectrum and n = 2 results in a 1D sweep of the illumination source (in addition to each pixel being an IF spectrum), n = 3 is obtained when using a 2D or 1D sweep of illumination with an RF frequency sweep, and n = 4 results in using a 2D illumination sweep with an RF frequency sweep. All or part of the following parameters can be changed.

Sweep reception (ie, “video”) is the main dimension of the frame.
• Sweep transmission (ie “radiation”)
RF frequency sweep IF frequency bin (read using parallel receive signal processor)
• Polarization sweep • Antenna beam pattern sweep (ie “depth of focus”)

  Such a high-dimensional data structure is regarded as a “frame”. The following analysis relies on this integrated representation “frame”.

  By starting from this (2 + n) D frame of composite data, each "pixel function" or "cost function" where each pixel is applied to all additional subsets or to a subset of data found in all dimensions A set of filters is provided that produces a scalar, positive, limited 2D image given by There is an intuitive “cost function” set that is grouped into function groups.

  1. Thus, the (2 + n) D frame generator fills the relevant locations in the frame memory 610 with spectral information for each pixel. Each spectrum is determined over a time window of a predetermined number of frames, and this time frame slides in time by one frame time to determine the spectrum of the next frame. The frame memory is accessed by a number of related devices, as described below. Assuming that all motion periods are slower than the frame interval,

  1. The IF correlator 615 looks for Doppler shifts in the IF that will indicate the motion of the object in the beam axis. This includes auto-correlation with the spectrum of the same pixel from the previous frame, thereby determining the frequency of the peak in the spectrum and whether this peak is moving with the frequency characteristics of the content flutter Decide whether or not. This provides an output that is sent through the adaptive filter 635 to form a false color once all the pixels have been processed and to show video areas with a common characteristic flutter frequency.

  2. Following established video filtering practices, spatial correlation is performed by the pixel correlator 620 using a 64 × 64 grid of pixels moving around the current pixel, which in this case is It is a spatial filtering of the spectrum, not just amplitude information for each pixel. Once all the pixels have been processed, adaptive filter 640 then creates another false color image to show the image area with the common characteristic spatial variation in the spectrum. This data field thus allows comparison between adjacent pixels, as the pixel projection area provides information about the vicinity of each pixel or pixel group of the frame. Such data is necessary for object extraction similar to the visual cortex in the human visual system that gives contour lines and the nickname “retina”.

  3. A temporal correlation is provided by the frame correlator 625, which looks for temporal repetitions or patterns as the spectrum of a given pixel changes over time. Thereafter, once all the pixels have been processed, adaptive filter 645 forms another false color image to indicate the image area having a common and characteristic temporal change in spectrum. This is summarized as a “memory” -like function that results in a typical duration found as pixel correlation time at each pixel. This filter provides speckle intensity and periodicity and is usually evaluated using spectral and pseudospectral methods. The pseudo-spectral content of the pixel therefore results in leaf flutter and minute movement signals. Such an operation requires a correlation between the previous frame and the actual one that yields the nickname “memory” for this unit. The temporal change seen when comparing various frames over time is caused by speckle, which tends to change at a speed specific to an object at a given distance and at a certain speed. As a result, a correlation peak is generated within a predetermined range of the time interval. Another cause of the time-dependent change is interference fringes from a thin layer such as a leaf that flutters in the wind in a given period. Other correlations are performed by the outer frame correlator 630 and summarized by the term “3D camera”, if the transmitter achieves this, different submillimeter bands, different parts of the IF range, different illumination depths or directions. And different focal positions can be used to include correlation between frames. Once all the pixels have been processed, another false color image is then formed by the adaptive filter 650, showing an area of the image with a common characteristic change in spectrum along with all of these factors.

  In FIG. 13, part B of this possible implementation of the video processor is shown. This figure shows various methods using various false color 2D images. Many false color images are supplied to the contour line extraction unit 665 via the weight function unit 660. The weighted false color video is overlaid or combined and applied to output a 2D video with enhanced threshold or other filtering, or simply provides metadata to add to the video. For example, for contour extraction, the largest weight can be given to the “retinal” video, some weight can be given to the Doppler video, and almost no weight is given to the memory and 3-D video. The process applied by contour extraction can be made dependent on input from a content feature library, which is known as a template that allows filling gaps and identifying shapes, for example. A shape can be provided.

  The region extraction portion 670 can operate in a similar manner, but can have a large weight from the topler image, eg, to identify an object such as a vehicle, and for example surface changes and Information from the content feature library can be used to identify the shape. The environmental parameter extraction unit 675 has an appropriate weight for identifying, for example, rain, canceling it in other images, or identifying ice on the road surface.

  In addition, a distance extractor 680 is shown, which is a different method of determining distances such as the size of the object that can be identified, different focal positions of the received beam to determine where the focus is sharpest. Combines information from different methods of ranging, by measuring the time of flight by means of correlation with noise modulation applied to the transmitter and detectable in the received signal, comparing the responses in be able to. The noise modulation can be randomized by jumping between different submillimeter wave bands. This is useful, for example, to avoid interference between similar radars on different vehicles in traffic jams.

  On the other hand, the frequency content of the IF channel can be deliberately manipulated in some embodiments by applying other low frequency sources (eg, ultrasound) to the target, and as a result, A sideband on the RF signal that is visible as a frequency shift in the IF channel is generated. Such an embodiment provides an analysis of structural vibrations, for example, to detect hidden cracks. The use of coherent demodulation provides better depth resolution and substantially provides holographic data. This allows the reconstruction of all RF modulations.

  By using these correlations, more information can be generated than simply adding false collars. For example, fast-changing items (leaves, shrubs) produce “white noise” in the color image, and other objects with a stable phase relationship that are repeated over one area of the object are color saturation that indicates their phase stability. Has a stable and unchanging color.

  Pair false images that have been acquired within the correlation time (at which time the object has no time to change) by false color pairing, and pair the acquired partial images at a larger correlation time. This extracts “persistence” or “correlation time” parameters that are specific to the particular object being imaged.

  For automotive radar applications, it is useful to provide images that are suitable for computer-based pattern extraction and that do not need to be optimized for the observer. Thus, the concept of “false color” (which has been used to present possibilities) is not limited to the three primary colors, but is a series of ( Potentially> 10) to correlate different partial videos.

FIG. 14: Alternative Embodiment Using Offline IF Processing and Video Processing This figure shows parts similar to those in FIG. 4 and refers to the corresponding description above. In this case, the IF and video processing part is replaced by the data acquisition unit 700, and the data acquisition unit 700 supplies data to the data storage unit 710. The offline video processing unit 755 executes functions similar to the functions of the IF processing and video processing unit in FIG.

FIGS. 15 and 16: Beam shaping and scanning for an embodiment of a radar system FIG. 15 shows a mechanical beam scanning for a transmitter. This can be replaced by electrical beam scanning. A rotating plane mirror 810 is provided on the transmit beam path following the first off-axis parabolic mirror 800 to provide an x-arrangement (also known as phi deflection in spherical coordinates). This is followed by a wave cylindrical mirror 820 for providing y-deflection (also known as theta-deflection in spherical coordinates), followed by a fixed secondary parabolic mirror 830 for providing an elliptical beam. is there. The timing control unit 840 is provided to control the moving part. Focus control is provided by the first mirror. If multiple beams are provided, they can be phase linked to have the effect of a single variable beam. A typical beam diameter is about 10 cm.

  FIG. 16 shows a similar part for receive beam scanning. The first element on the beam path is a fixed secondary parabola 890, followed by a wave plane mirror 870 for providing y deflection. This is followed by a rotating mirror 860 to provide x deflection followed by a movable off-axis parabolic mirror 850 to provide focus control. In addition, a timing controller 900 is provided to control and synchronize the movement of the three moving parts.

Figures 17 to 23: How content flutter and speckle effects provide detectable information in the spectrum of the received signal . Figure 17 shows the IF of a given pixel when directed to a given object. It is a graph which shows an example of the time characteristic (signature) of a signal. Temporal features (signatures) include sequences where the IF amplitude changes in a pseudo periodic manner generated by changes in the run time of the radar signal. Since there is a standing wave between the transmitter and receiver antennas, a change in half-wave run time changes the standing wave from a constructive interference case to a destructive interference case. All objects that have a constant velocity with respect to the radar system will cause periodic speckle. An object with a more random velocity produces a pseudo-periodic speckle, where the temporal length of the pseudo-periodic sequence depends on the rate of rate change. The first part of the signature shows the effect of the slower speed object and the second part shows the effect of the faster speed object.

  FIG. 18 shows a graph of the frequency content IF signal for a given pixel when directed to a selected type of target showing a window of frequencies used for object classification. Window 1 is used for background removal, window 2 is used for identification of the target of interest, and window 3 serves for plant removal purposes.

  The thick solid curve shows the frequency content of the IF obtained for a rapidly moving target (eg, a plant). A thick dotted curve shows the frequency content of the IF obtained for a slowly moving target (eg, a pedestrian). This represents the spectrum of the time signal shown in FIG. The thin dotted curve represents the IF frequency content obtained for a signal returning from a fixed background. This represents the frequency spectrum of the time signal shown in FIG.

Table 1 shows a table of average values of absolute values of FFT acquired on the window shown in FIG. The value in parentheses relates to the content in window 1.

This table will be described below.
Of course, the object that moves at high speed has the largest contribution in the window 3. There is nothing in the background, and objects that move slowly are quite small there. If relative values are taken into account, windows 2 and 3 show similar behavior. For plants and for pedestrians, windows 1 and 2 behave similarly. As for the background, you can't see much in windows 2 and 3.

  FIG. 21 shows a plot of the raw amplitude data for all pixels in the frame, for comparison with the processed video shown in FIGS. This is the average of the values obtained in the frequency window for each pixel that can be plotted against all pixels to obtain a processed image as shown in FIGS. FIG. 22 shows a processed video corresponding to window 3, which highlights a fast moving plant as described above with respect to FIG. FIG. 23 shows a processed image corresponding to the difference between windows 2 and 3, which emphasizes pedestrians and minimizes fast moving plants and static background. . This is because the pedestrian has the largest difference between windows 2 and 3 as shown in FIG.

  FIG. 22 represents the amplitude in the fastest frequency window (# 3) in the same scenario as in FIG. FIG. 23 shows the difference between the slow (# 2) and fast frequency window (# 3) weighted by appropriate constants. All plots are contour plots normalized between the maximum and minimum values of the plot. One crossing of the contour line corresponds to a 1% change in the value span.

  Other processed images can be obtained to distinguish features. For example, an enhanced “slow” component image can be obtained by multiplying the intensity by factor 3 or other factors.

FIGS. 24 and 25: Dielectric thin film layer causing Fabry-Perot effect in RF FIGS. 24 and 25 show how the presence of a dielectric thin film layer affects the RF frequency dependence of the radar signal. These figures show a plot of amplitude versus RF frequency to show the Fabry-Perot effect in metal at RF. In these figures, the solid line indicates the real part of the reflected power. The dotted line indicates the imaginary part of the reflected power.

  FIG. 24 shows glazing reflection from an uncoated metal surface (at a low incidence angle, approximately parallel to the surface). FIG. 25 shows a glazing reflection from a similar metal surface that is coated with a 2 mm thick plastic layer so that the beam path length through the layer is about 60 mm. From the comparison of the graphs, it can be seen that the responses in FIG. 25 differ mainly in the large and periodic fluctuations on the left hand side. This variation corresponds to constructive and destructive interference that varies with frequency. This can therefore be used to determine whether an object has a surface layer of a predetermined thickness or whether the thickness is known or whether information about the angle of incidence can be derived. The frequency range in which periodic variations appear can change due to second order effects, but the presence of periodic variations is typical of the surface layer.

  The signal processing apparatus of the present invention can be realized as hardware, computer software, or a combination thereof. The signal processor may be a general purpose processor, embedded processor, application specific integrated circuit (ASIC), field programmable gate array (FPGA) or other programmable logic device, discrete gate or transistor logic, discrete hardware component, or All combinations designed to achieve the functions described herein can be included. The signal processing apparatus further includes a combination of computing devices, for example, a combination of an FPGA and a microprocessor, a plurality of microprocessors, one or more microprocessors coupled to the FPGA, or other such configurations. Can do.

  The present invention further includes a computer program product that provides the functionality of all methods according to the present invention when executed on a computing device. When executed on a processing engine, the software according to the present invention may include a code segment that provides a method of using a sub-millimeter wavelength active radar system. This method is used to radiate the field of view with a transmitter, the software receives the signal from the field of view, and the downconverted signal is periodic with phase and amplitude information that depends on the content in the field of view. It is adapted to allow the received signal to be down-converted coherently using a phase-locked demodulator to have a component. The software, when executed, is adapted to further process the downconverted signal to identify information about the content from the phase and amplitude information in the signal.

  This signal can have a periodic component with phase and amplitude information that depends on the content in the radar system's field of view, and the software uses a local oscillator signal that is phase locked to the transmitter of the radar system. It is adapted to downconvert the signal and further identify information about the content from the phase and amplitude information in the signal.

  The software may be adapted to identify information by applying one or more operators to the received signal to extract the spatial or temporal characteristics of the spectrum of the received signal.

  Other variations are possible within the scope of the claims of the present invention.

Claims (12)

In a signal processing apparatus for a submillimeter wave active radar system having a field of view,
The signal processing device is configured to process a signal received from the field of view and downconverted by the radar system;
The down-converted signal corresponding to a predetermined pixel, and a time-varying amplitude component and temporally varying phase component, the content in the predetermined pixel of the amplitude and phase components together in the field of view have a periodic component that is dependent on the down-converted signal comprises a separate signal received at a plurality of different sub-millimeter-wave frequencies, and further, the signal processing apparatus, the down-converted signal corresponding to the different sub-millimeter wave frequencies Are configured to be analyzed separately, and
The signal processing device is configured to identify information about the content from the periodic component;
Information identified from the periodic component includes content flutter, temporal and spatial speckle patterns specific to the content, and interference effects from multiple reflections in a thin layer in the content resulting in radio frequency dependent periodicity; A signal processing device comprising effects from at least one selected from.   2. The signal processing device according to claim 1, wherein the down-converted signal is an IF signal, and further, the signal processing device is configured to separately analyze two or more regions of the IF spectrum. , Signal processing device. 3. The signal processing device according to claim 1, wherein the signal processing device is configured to detect and classify an object in the content using the identified information. The signal processing apparatus according to any one of claims 1 to 3, having a portion for relating a portion of the received signal to a corresponding position in the field of view, the signal processing device. The signal processing apparatus according to any one of claims 1 to 4, in order to extract the spatial or temporal characteristics of components of the down-converted signal, one or more of the components of the down-converted signal A signal processing apparatus comprising a portion for applying an operator to identify the information. 6. The signal processing apparatus according to claim 5 , wherein the operator has one or more of the following characteristics depending on a submillimeter wave band, a radiation direction, a focal position of a receiver, or polarization. A signal processing apparatus comprising one operator for extraction. In a submillimeter wave radar system having a transmitter for emitting a field of view, a receiver for receiving a signal from the field of view, and a demodulator, the demodulator is phase locked to the transmitter; The said system is a submillimeter wave radar system which has the signal processing apparatus of any one of Claims 1 thru | or 5 . A vehicle having a radar system according to claim 6 and an output system for using the identified information. In a method using a submillimeter wave active radar system, the method comprises:
Radiate the field of view with a transmitter,
Receiving a signal from the field of view;
Downconverted signal corresponding to a predetermined pixel of said field of view, and the amplitude component time-varying, and a phase component which varies with time, the amplitude and phase components is predetermined in the field of view both Using a demodulator phase-locked to the transmitter so that the down-converted signal includes separate signals received at different submillimeter frequencies and having a periodic component that depends on the content in the pixel Downconvert the signal coherently ,
Analyzing each of the downconverted signals corresponding to the different submillimeter frequencies separately and further processing the downconverted signal to identify information about the content from the periodic component;
The information identified from the periodic components includes content flutter, content-specific temporal and spatial speckle patterns, and interference effects from multiple reflections in a thin layer in the content, and a frequency-dependent period. A method comprising at least one effect selected from an interference effect that produces sex. In a method for processing a signal from a receiver of a submillimeter wave radar system,
The signal depends on content in the field of view of the radar system, and the method further comprises:
Downconverted signal corresponding to a predetermined pixel in said field of view having an amplitude varying components temporally, and a phase component which varies with time, the amplitude and phase components together has a periodic component And downconverting the signal using a demodulator phase locked to the radar system transmitter so that the downconverted signal includes separate signals received at different submillimeter frequencies ,
Analyzing the downconverted signals corresponding to the different submillimeter frequencies separately;
Identifying each piece of content information from the periodic component,
The information identified from the periodic components includes content flutter, content-specific temporal and spatial speckle patterns, and interference effects from multiple reflections in a thin layer in the content, and a frequency-dependent period. A method comprising at least one effect selected from an interference effect that produces sex. The method of claim 10 , further comprising identifying the information by applying one or more operators to the received signal to extract a spatial or temporal characteristic of the spectrum of the received signal. A method comprising the steps of: A computer readable medium having a stored program that, when executed by a computer, causes the computer to perform the steps of claim 10 or 11 .

Images